Abstract
Visual Working Memory (VWM) is commonly characterized using a change detection task (Luck & Vogel, 1997), but change detection errors reveal little about trial-by-trial performance. That is, errors produced on high-success trials (many items remembered) and low-success trials (no items remembered) are indistinguishable. Here, we employ a discrete whole-report task to precisely track trial-by-trial VWM performance. In the discrete whole-report paradigm, participants view a briefly presented array of colored squares. After a brief retention interval, colored grids appear at the locations of all items. Participants report the color of all items by clicking the color in each grid that corresponds with the remembered color. In Experiment 1, participants completed a change detection task (set sizes 2-6) and a whole-report task (30 trials each of set sizes 2-6). In Experiment 2, participants completed a change detection task (set sizes 4, 6, & 8) and 300 trials of whole-report (set size 6 only). The mean number of correct items in the whole-report task strongly corresponded with change detection capacity (Exp 1: R2 = .52, p = <.001; Exp 2: R2 = .43, p = <.001 ). Additionally, proportions of low-performance (<3) and high-performance (> 3) whole-report trials predicted VWM capacity. However, the majority of both high- and low-capacity subjects correctly identified a modal number of 3 items, and the proportion of whole-report trials with 3 items correct did not predict individual differences in VWM capacity. The present results reveal that precise characterization of high- and low-success trials has important and surprising implications for models of VWM. In particular, these results provide evidence that stable individual differences in VWM capacity should be be conceptualized as differences in successfully utilizing memory resources rather than as differences in total resources available.
Meeting abstract presented at VSS 2014